Geographic origin discrimination of pork from different Chinese regions using mineral elements analysis assisted by machine learning techniques

Underlying characteristic aroma of white tea from diverse geographical origins and its prediction

BACKGROUND

White tea, an agriculturally distinctive product, exhibits significant aroma variations across different regions. Nevertheless, the mechanisms driving these differences, and distinguishing methods suitable for specific origins, have been scarcely reported. In this study, we analyzed the aroma characteristics and volatile components of 100 white tea samples from ten regions, utilizing sensory evaluation, headspace solid-phase microextraction-gas chromatography-mass spectrometry and chemometrics, then established a discrimination model.

RESULTS

A total of 66 volatile compounds were identified, with alcohols and esters being the most important. Linalool and geranyl alcohol were common and relatively abundant volatile compounds across all ten regions, significantly contributing to the aroma characteristics of white tea. The relative content of volatile compounds differed notably across regions, where 33 key compounds, including (E)-2-phenylbut-2-enal and methyl 2,5-octadecadiynoate, were crucial for regional prediction. Employing machine learning algorithms, such as random forest and support vector machine for regional prediction, yielded accuracies of 93.33% and 90.00%, respectively.

CONCLUSION

This study unveils new insights into aroma variation in white tea across different origins, proposing an innovative way of origin determination. © 2025 Society of Chemical Industry.

Constructing an origin discrimination model of japonica rice in Heilongjiang Province based on confocal microscopy Raman spectroscopy technology

An origin discrimination model for rice from five production regions in Heilongjiang Province was constructed based on the combination of confocal microscopy Raman spectroscopy and chemometrics. A total of 150 field rice samples were collected from the Fangzheng, Chahayang, Jiansanjiang, Xiangshui, and Wuchang production areas. The optimal sample processing conditions, instrument parameter settings, and spectrum acquisition techniques were identified by investigating the influencing factor. The Raman spectra of milled rice within the range of 100-3200 cm-1 were selected as the raw data, and the optimal preprocessing method combination consisting of normalization, Savitzky-Golay smoothing, and multivariate scatter correction was identified. Subsequently, the competitive adaptive reweighted sampling and discrete binary particle swarm optimization algorithms were employed to optimize the feature wavelength selection, resulting in the screening of 226 and 1899 feature wavelength variables, respectively. Using the full Raman spectrum data and feature wavelength data as inputs, partial least squares discriminant analysis, support vector machine and extreme learning machine origin discrimination models were constructed. The results indicated that the BPSO-GA-SVM model exhibited the best predictive ability, achieving a testing set accuracy of 86.67%.

Recent advances of machine learning in the geographical origin traceability of food and agro-products: A review

The geographical origin traceability of food and agro-products has been attracted worldwide. Especially with the rise of machine learning (ML) technology, it provides cutting-edge solutions to erstwhile intractable issues to identify the origin of food and agro-products. By utilizing advanced algorithms, ML can extract feature information of food and agro-products that is closely related to origin and, more accurately, identify and trace their origins, which is of great significance to the entire food and agriculture industry. This paper provides a comprehensive overview of the state-of-the-art applications of ML in the geographical origin traceability of food and agro-products. First, commonly used ML methods are summarized. The paper then outlines the whole process of preparation for modeling, model training as well as model evaluation for building traceability models-based ML. Finally, recent applications of ML combined with different traceability techniques in the field of food and agro-products are revisited. Although ML has made many achievements in solving the geographical origin traceability problem of food and agro-products, it still has great development potential. For example, the application of ML is yet insufficient in the geographical origin traceability using DNA or computer vision techniques. The ability of ML to predict the geographical origin of food and agro-products can be further improved, for example, by increasing model interpretability, incorporating data fusion strategies, and others.

Comparison and chemometrics analysis of phenolic compounds and mineral elements in Artemisia Argyi Folium from different geographical origins

The quality of Artemisia Argyi Folium (AAF), a traditional Chinese food ingredient, is intrinsically linked to its geographical origin, which this study explores through phenolic compounds and mineral elements. The contents of 17 phenols and 18 minerals differed significantly between geographically distinct samples according to UHPLC and ICP-MS, respectively. Chemometrics indicated that a supervised model, orthogonal partial least squares discriminant analysis (OPLS-DA), outperformed unsupervised methods at classifying AAF samples by their origins. Phenols were more effective at distinguishing samples from seven provinces, while minerals were adept at differentiating samples from the Dabie Mountain region (three provinces) and those from four other provinces. Six phenols and 10 minerals were important variables for discrimination. Complex correlations were observed between the contents of various phenols and minerals in AAF, with minerals possibly affecting the accumulation of phenols. This study provides an approach for distinguishing geographically distinct AAF samples and determining their geographical origins.

Differentiating Pond-Intensive, Paddy-Ecologically, and Free-Range Cultured Crayfish (Procambarus clarkii) Using Stable Isotope and Multi-Element Analysis Coupled with Chemometrics

The farming pattern of crayfish significantly impacts their quality, safety, and nutrition. Typically, green and ecologically friendly products command higher economic value and market competitiveness. Consequently, intensive farming methods are frequently employed in an attempt to replace these environmentally friendly products, leading to potential instances of commercial fraud. In this study, stable isotope and multi-element analysis were utilized in conjunction with multivariate modeling to differentiate between pond-intensive, paddy-ecologically, and free-range cultured crayfish. The four stable isotope ratios of carbon, nitrogen, hydrogen, and oxygen (δ13C, δ15N, δ2H, δ18O) and 20 elements from 88 crayfish samples and their feeds were determined for variance analysis and correlation analysis. To identify and differentiate three different farming pattern crayfish, unsupervised methods such as hierarchical cluster analysis (HCA) and principal component analysis (PCA) were used, as well as supervised multivariate modeling, specifically partial least squares discriminant analysis (PLS-DA). The HCA and PCA exhibited limited effectiveness in classifying the farming pattern of crayfish, whereas the PLS-DA demonstrated a more robust performance with a predictive accuracy of 90.8%. Additionally, variables such as δ13C, δ15N, δ2H, Mn, and Co exhibited relatively higher contributions in the PLS-DA model, with a variable influence on projection (VIP) greater than 1. This study is the first attempt to use stable isotope and multi-element analysis to distinguish crayfish under three farming patterns. It holds promising potential as an effective strategy for crayfish authentication.

Combined quantitative lipidomics and back-propagation neural network approach to discriminate the breed and part source of lamb

The study successfully utilized an analytical approach that combined quantitative lipidomics with back-propagation neural networks to identify breed and part source of lamb using small-scale samples. 1230 molecules across 29 lipid classes were identified in longissimus dorsi and knuckle meat of both Tan sheep and Bahan crossbreed sheep. Applying multivariate statistical methods, 12 and 7 lipid molecules were identified as potential markers for breed and part identification, respectively. Stepwise linear discriminant analysis was applied to select 3 and 4 lipid molecules, respectively, for discriminating lamb breed and part sources, achieving correct rates of discrimination of 100 % and 95 %. Additionally, back-propagation neural network proved to be a superior method for identifying sources of lamb meat compared to other machine learning approaches. These findings indicate that integrating lipidomics with back-propagation neural network approach can provide an effective strategy to trace and certify lamb products, ensuring their quality and protecting consumer rights.

Fatty Acid Profiling as a Tool for Fostering the Traceability of the Halophyte Plant Salicornia ramosissima and Contributing to Its Nutritional Valorization

Salicornia ramosissima, commonly known as glasswort or sea asparagus, is a halophyte plant cultivated for human consumption that is often referred to as a sea vegetable rich in health-promoting n-3 fatty acids (FAs). Yet, the effect of abiotic conditions, such as salinity and temperature, on the FA profile of S. ramosissima remains largely unknown. These factors can potentially shape its nutritional composition and yield unique fatty acid signatures that can reveal its geographical origin. In this context, samples of S. ramosissima were collected from four different locations along the coastline of mainland Portugal and their FAs were profiled through gas chromatography-mass spectrometry. The lipid extracts displayed a high content of essential FAs, such as 18:2n-6 and 18:3n-3. In addition to an epoxide fatty acid exclusively identified in samples from the Mondego estuary, the relative abundance of FAs varied between origin sites, revealing that FA profiles can be used as site-specific lipid fingerprints. This study highlights the role of abiotic conditions on the nutritional profile of S. ramosissima and establishes FA profiling as a potential avenue to trace the geographic origin of this halophyte plant. Overall, the present approach can make origin certification possible, safeguard quality, and enhance consumers' trust in novel foods.

Towards Verifying the Imported Soybeans of China Using Stable Isotope and Elemental Analysis Coupled with Chemometrics

Verifying the geographical origin of soybeans (Glycine max [Linn.] Merr.) is a major challenge as there is little available information regarding non-parametric statistical origin approaches for Chinese domestic and imported soybeans. Commercially procured soybean samples from China (n = 33) and soybeans imported from Brazil (n = 90), the United States of America (n = 6), and Argentina (n = 27) were collected to characterize different producing origins using stable isotopes (δ2H, δ18O, δ15N, δ13C, and δ34S), non-metallic element content (% N, % C, and % S), and 23 mineral elements. Chemometric techniques such as principal component analysis (PCA), linear discriminant analysis (LDA), and BP-artificial neural network (BP-ANN) were applied to classify each origin profile. The feasibility of stable isotopes and elemental analysis combined with chemometrics as a discrimination tool to determine the geographical origin of soybeans was evaluated, and origin traceability models were developed. A PCA model indicated that origin discriminant separation was possible between the four soybean origins. Soybean mineral element content was found to be more indicative of origin than stable isotopes or non-metallic element contents. A comparison of two chemometric discriminant models, LDA and BP-ANN, showed both achieved an overall accuracy of 100% for testing and training sets when using a combined isotope and elemental approach. Our findings elucidate the importance of a combined approach in developing a reliable origin labeling method for domestic and imported soybeans in China.

Effects of different floral periods and environmental factors on royal jelly identification by stable isotopes and machine learning analyses during non-migratory beekeeping

It is crucial to monitor the authenticity of royal jelly (RJ) because the qualities of RJs produced by different floral periods vary substantially. In the context of non-migratory beekeeping, this study aims to identify rape RJ (RRJ), chaste RJ (CRJ), and sesame RJ (SRJ) based on δ13C, δ15N, δ2H, and δ18O combined with machine learning and to evaluate environmental effect factors. The results showed that δ13C (-27.62‰ ± 0.24‰), δ15N (2.88‰ ± 0.85‰), and δ18O (28.02‰ ± 1.30‰) of RRJ were significantly different from other RJs. The δ13C, δ2H, and δ18O in CRJ and SRJ were strongly correlated with temperature and precipitation, suggesting that these isotopes are influenced by environmental elements such as sunlight and rainfall. In addition, the artificial neural network (ANN) model was superior to the random forest (RF) model in terms of accuracy, sensitivity, and specificity. This study revealed that combining stable isotopes with ANN models and the unique correlation between stable isotopes and environmental factors could provide promising ideas for monitoring the authenticity of RJ.

Elemental profiles of swine tissues as descriptors for the traceability of value-added Italian heavy pig production chains

The increasing demand for reliable traceability tools in the meat supply chain has prompted the exploration of innovative approaches that meet stringent quality standards. In this work, 57 elements were quantified by inductively coupled plasma mass spectrometry and direct mercury analysis in 80 muscle and 80 liver samples of Italian heavy pigs to investigate the potential of new tools based on multi-elemental profiles in supporting value-added meat supply chains. Samples from three groups of animals belonging to the protected designation of origin (PDO) Parma Ham circuit (conventionally raised; raised with genetically modified organism (GMO)-free feeds; raised with GMO-free feeds plus the supplementation of omega-3 polyunsaturated fatty acids (n-3 PUFA)) and a fourth group of samples from animals not compliant with the PDO Parma Ham production process were analyzed. Hierarchical cluster analysis allowed for the identification of three macro-clusters of liver or muscle samples, highlighting some inhomogeneities among the target groups. Following SIMCA analysis, better classification models were obtained by using liver elemental profiles (95% correct classification rate), with the highest classification accuracy observed for GMO-free livers (100%). The elements contributing the most to the separation of livers by class membership were La, Ce, and Pb for conventional, Li, Cr, Fe, As, and Sr for GMO-free + n-3 PUFA, and Lu for non-PDO samples. Given these findings, the analysis of the elemental profiles of pig tissues can be regarded as a promising method to confirm the declared pig meat label attributes, deter potential complex fraud, and support meat traceability systems.

Comparative antioxidant activity and untargeted metabolomic analyses of cherry extracts of two Chinese cherry species based on UPLC-QTOF/MS and machine learning algorithms

P. pseudocerasus and P. tomentosa are the two native Chinese cherry species of high economic and ornamental worths. Little is known about the metabolic information of P. pseudocerasus and P. tomentosa. Effective means are lacking for distinguishing these two similar species. In this study, the differences in total phenolic content (TPC), total flavonoid content (TFC), and in vitro antioxidant activities in 21 batches of two species of cherries were compared. A comparative UPLC-QTOF/MS-based metabolomics coupled with three machine learning algorithms was established for differentiating the cherry species. The results demonstrated that P. tomentosa had higher TPC and TFC with average content differences of 12.07 times and 39.30 times, respectively, and depicted better antioxidant activity. Total of 104 differential compounds were identified by UPLC-QTOF/MS metabolomics. The major differential compounds were flavonoids, organooxygen compounds, and cinnamic acids and derivatives. Correlation analysis revealed differences in flavonoids content such as procyanidin B1 or isomer and (Epi)catechin. They could be responsible for differences in antioxidant activities between the two species. Among three machine learning algorithms, the prediction accuracy of support vector machine (SVM) was 85.7%, and those of random forest (RF) and back propagation neural network (BPNN) were 100%. BPNN exhibited better classification performance and higher prediction rate for all testing set samples than those of RF. The study herein found that P. tomentosa had higher nutritional value and biological functions, and thus considered for usage in health products. Machine models based on untargeted metabolomics can be effective tools for distinguishing these two species.

Towards a Standardized Approach for the Geographical Traceability of Plant Foods Using Inductively Coupled Plasma Mass Spectrometry (ICP-MS) and Principal Component Analysis (PCA)

This paper presents a systematic literature review focused on the use of inductively coupled plasma mass spectrometry (ICP-MS) combined with PCA, a multivariate technique, for determining the geographical origin of plant foods. Recent studies selected and applied the ICP-MS analytical method and PCA in plant food geographical traceability. The collected results from many previous studies indicate that ICP-MS with PCA is a useful tool and is widely used for authenticating and certifying the geographic origin of plant food. The review encourages scientists and managers to discuss the possibility of introducing an international standard for plant food traceability using ICP-MS combined with PCA. The use of a standard method will reduce the time and cost of analysis and improve the efficiency of trade and circulation of goods. Furthermore, the main steps needed to establish the standard for this traceability method are reported, including the development of guidelines and quality control measures, which play a pivotal role in providing authentic product information through each stage of production, processing, and distribution for consumers and authority agencies. This might be the basis for establishing the standards for examination and controlling the quality of foods in the markets, ensuring safety for consumers.

Metabolomics integrated with machine learning to discriminate the geographic origin of Rougui Wuyi rock tea

The geographic origin of agri-food products contributes greatly to their quality and market value. Here, we developed a robust method combining metabolomics and machine learning (ML) to authenticate the geographic origin of Wuyi rock tea, a premium oolong tea. The volatiles of 333 tea samples (174 from the core region and 159 from the non-core region) were profiled using gas chromatography time-of-flight mass spectrometry and a series of ML algorithms were tested. Wuyi rock tea from the two regions featured distinct aroma profiles. Multilayer Perceptron achieved the best performance with an average accuracy of 92.7% on the training data using 176 volatile features. The model was benchmarked with two independent test sets, showing over 90% accuracy. Gradient Boosting algorithm yielded the best accuracy (89.6%) when using only 30 volatile features. The proposed methodology holds great promise for its broader applications in identifying the geographic origins of other valuable agri-food products.

A New and Effective Method to Trace Tibetan Chicken by Amino Acid Profiling

As a "rare bird on the plateau", the Tibetan chicken is rich in nutrition and has high medicinal value. In order to quickly and effectively identify the source of food safety problems and to label fraud regarding this animal, it is necessary to identify the geographical traceability of the Tibetan chicken. In this study, Tibetan chicken samples from four different cities in Tibet, China were analyzed. The amino acid profiles of Tibetan chicken samples were characterized and further subjected to chemometric analyses, including orthogonal least squares discriminant analysis, hierarchical cluster analysis, and linear discriminant analysis. The original discrimination rate was 94.4%, and the cross-validation rate was 93.3%. Moreover, the correlation between amino acid concentrations and altitudes in Tibetan chicken was studied. With the increase in altitude, all amino acid contents showed a normal distribution. For the first time, amino acid profiling has been comprehensively applied to trace the origin of plateau animal food with satisfactory accuracy.

Growth performance, meat quality, and blood characteristics of finisher crossbred pigs fed diets supplemented with different levels of green tea (Camellia sinensis) by-products

Background and Aim

Dietary supplementation with green tea by-product shows special effects on animal parameters. This study aimed to assess the effects of green tea by-products (GTBP) in the diet on some blood parameters, growth performance, and carcass characteristics of finishing pigs and on meat quality, and nutritional composition of pork.

Materials and Methods

One hundred and sixty crossbred pigs with an initial body weight of 65.15 ± 0.38 kg, were distributed into four dietary treatments, with four replicates of 10 pigs each. The dietary treatments were a basal diet (control diet, CON), and three experimental diets (GTBP8, GTBP16, and GTBP24) based on the CON diet supplemented with GTBP at 8, 16, and 24 g/kg of feed. The studied parameters were examined during the experimental period of 10 weeks.

Results

No statistical differences in average daily feed intake, average daily gain, and feed conversion ratio were observed between the diet treatments (p > 0.05). Backfat thickness decreased (linear, p < 0.05) according to the GTBP levels but no other carcass parameters. Meat quality was not influenced by the GTBP levels (p > 0.05). However, pigs fed with GTBP had a decrease in cholesterol content and an increase in crude protein and total omega-3 content of pork compared to the CON diet (p < 0.05). Moreover, dietary supplementation with GTBP significantly decreased plasma cholesterol (p < 0.05), and trends for the decrease in low-density lipoprotein cholesterol and urea nitrogen were observed (linear, p = 0.08).

Conclusion

Up to 24 g/kg GTBP in the diet for finishing pigs does not impair animal performance and makes carcass leaner with softer meat as well as positive effects on cholesterol and fatty acid metabolism. Further experiments are needed to determine the optimal levels of GTBP addition in finishing pig diet to produce higher meat quality.

The Effects of Dietary Inclusion of Mulberry Leaf Powder on Growth Performance, Carcass Traits and Meat Quality of Tibetan Pigs

This research was conducted to study the effects of dietary inclusion of mulberry leaf powder (MLP) on growth performance, meat quality, antioxidant activity, and carcass traits of Tibetan pigs. Eighteen Tibetan pigs (33.8 ± 1.1 kg) were assigned to two treatment groups randomly and received either the control diet (CON) or a basal diet containing 8% MLP (MLP) for two months. After the two-month feeding trial, the MLP group showed lower backfat thickness while a higher lean percentage. Compared with CON pigs, MLP pigs had higher serum CAT activity. In addition, dietary MLP supplementation significantly decreased the muscle shear force. Muscle fiber morphology analysis showed that MLP pigs had larger muscle fiber density while smaller muscle fiber cross-sectional area. Up-regulated gene expression of myosin heavy chain (MyHC)IIa was also observed in MLP pigs. These results indicate that the enhanced antioxidant activity and altered muscle fiber type and morphology appeared to contribute to the improvement of meat quality in Tibetan pigs fed diets containing MLP.

Handling multiblock data in wine authenticity by sequentially orthogonalized one class partial least squares

New approach to deal with food authentication by modelling methods based on data recorded from different sources is proposed and called OC-PLS, combines an orthogonalization step between the different data sets to eliminate redundant information followed by definition of an acceptance area for a target class by OC-PLS. The proposed method was evaluated in two case studies. The first study used a controlled scenario with simulated data. In the second case study, the approach was applied using UV-VIS and IR data, in order to differentiate Slovak Tokaj Selection wines of high quality from other lower market value wines from the Slovak Tokaj wine region. In both cases, better results were reached than when individual blocks of data were achieved. The proposed method proved to be effective in properly exploring common and distinct information in each data block. The best compromise between sensitivity and selectivity in the prediction step was achieved.

Multi-Element Analysis and Origin Discrimination of Panax notoginseng Based on Inductively Coupled Plasma Tandem Mass Spectrometry (ICP-MS/MS)

Panax notoginseng is an important functional health product, and has been used worldwide because of a wide range of pharmacological activities, of which the taproot is the main edible or medicinal part. However, the technologies for origin discrimination still need to be further studied. In this study, an ICP-MS/MS method for the accurate determination of 49 elements was established, whereby the instrumental detection limits (LODs) were between 0.0003 and 7.716 mg/kg, whereas the quantification limits (LOQs) were between 0.0011 and 25.7202 mg/kg, recovery of the method was in the range of 85.82% to 104.98%, and the relative standard deviations (RSDs) were lower than 10%. Based on the content of multi-element in P. notoginseng (total of 89 mixed samples), the discriminant models of origins and cultivation models were accurately determined by the neural networks (prediction accuracy was 0.9259 and area under ROC curve was 0.9750) and the support vector machine algorithm (both 1.0000), respectively. The discriminant models established in this study could be used to support transparency and traceability of supply chains of P. notoginseng and thus avoid the fraud of geographic identification.

Geographic Origin Discrimination of Millet Using Vis-NIR Spectroscopy Combined with Machine Learning Techniques

Millet is a primary food for people living in the dry and semi-dry regions and is dispersed within most parts of Europe, Africa, and Asian countries. As part of the European Union (EU) efforts to establish food originality, there is a global need to create Protected Geographical Indication (PGI) and Protected Designation of Origin (PDO) of crops and agricultural products to ensure the integrity of the food supply. In the present work, Visible and Near-Infrared Spectroscopy (Vis-NIR) combined with machine learning techniques was used to discriminate 16 millet varieties (n = 480) originating from various regions of China. Five different machine learning algorithms, namely, K-nearest neighbor (K-NN), Linear discriminant analysis (LDA), Logistic regression (LR), Random Forest (RF), and Support vector machine (SVM), were used to train the NIR spectra of these millet samples and to assess their discrimination performance. Visible cluster trends were obtained from the Principal Component Analysis (PCA) of the spectral data. Cross-validation was used to optimize the performance of the models. Overall, the F-Score values were as follows: SVM with 99.5%, accompanied by RF with 99.5%, LDA with 99.5%, K-NN with 99.1%, and LR with 98.8%. Both the linear and non-linear algorithms yielded positive results, but the non-linear models appear slightly better. The study revealed that applying Vis-NIR spectroscopy assisted by machine learning technique can be an essential tool for tracing the origins of millet, contributing to a safe authentication method in a quick, relatively cheap, and non-destructive way.

Combing machine learning and elemental profiling for geographical authentication of Chinese Geographical Indication (GI) rice

Identification of geographical origin is of great importance for protecting the authenticity of valuable agri-food products with designated origins. In this study, a robust and accurate analytical method that could authenticate the geographical origin of Geographical Indication (GI) products was developed. The method was based on elemental profiling using inductively coupled plasma mass spectrometry (ICP-MS) in combination with machine learning techniques for model building and feature selection. The method successfully predicted and classified six varieties of Chinese GI rice. The elemental profiles of 131 rice samples were determined, and two machine learning algorithms were implemented, support vector machines (SVM) and random forest (RF), together with the feature selection algorithm Relief. Prediction accuracy of 100% was achieved by both Relief-SVM and Relief-RF models, using only four elements (Al, B, Rb, and Na). The methodology and knowledge from this study could be used to develop reliable methods for tracing geographical origins and controlling fraudulent labeling of diverse high-value agri-food products.